THE AMERICAN MINERAI-OGIST, VOL. 46, SEPTEMBER-OCTOBER. 1961 PHASE TRANSFORMATIONS IN SILICA-ALUMINA MIXTURES AS EXAMINED BY CONTINUOUS X-RAY DIFFRACTION F. M. W.tsr,, R. E. Gnru, and R. B. Gnar',* Unioersity oj Ill in ois, Urba n a, I I I i nois. Assrnecr The {ormation of mullite from mixtures of different varieties of alumina and silica was studied by high-temperature continuous r-ray diffraction techniques and by diffraction after soaking at high temperatures for various lengths of time. ft is shown that the initial form of the alumina and silica has a large influence and that mole per cent variations in mix composition have little efiect on mullite development. The structural stability of component materials largely controls mullite nucieation, and the relationship between the formation of this mineral and preceding transformations in the normal thermal sequence of component alumina and silica materials is emphasized. The addition of mineralizing ions can promote mullitization. INrnooucrroN 'fhe equilibrium diagram for the system AlzOa-SiOz has been critically examined and re-evaluated by many persons since it was first determined by Bowen and Greig in 1924. The most recent revision is that of Aramaki and Roy (1960) and is based on data from more than 700 runs. The standard quenching technique has generally been used in studying silicate phase equilibria, and liquidus-solidus relationships at high temperatures have received primary consideration. The information to be presented herein also deals with high-temperature mineral transformations within the alumina-silica system; however, special consideration is given to the 900o-1500' C. temperature range and to the formation of mullite as it develops from synthetic mixtures of different varieties of alumina and silica. Insley and Ewell in 1935, and more recently West and Gray (1958) in examining reactions in artificial alumina-silica mixtures, suggested the influence which internal structures of materials could have in controlling thermal reactions within this temperature range. In the present study the development of mullite from various mixtures of alumina-silica materials has been investigated by both continuous heating r.-ray diffraction and r-ray difrra"ction after prolonged heating at elevated temperatures, and it will be shown that temperature and the structural nature of component materials are more important than mole per cent variations in the composition of the mixes in controlling the development of this mineral at elevated temperatures. * Present Address: General Dynamics Corporation, Fort Worth, Texas. t064
PHASE TRANSFORMATIONS IN Si'OZ-AI'ZOI MIXTARES 1065 Sarupr-Bs AND PRocEDURE The following varieties of alumina and silica were selected for firing: diaspore from Missouri; synthetic gibbsite, Merck and Co., Rahway, New Jersey; and corundum from Zoutpansberg District, Transvaal, Africa; also synthetic cristobalite; silicic acid, 100 mesh, A. R. grade, Mallinckrodt Chemical Works; and Ottawa quartz sand from near Ottawa, Illinois. The effect of mole per cent variations of both alumina and silica on mullite formation was investigated by *-raying continuously diasporecristobalite mixtures of Alzor'4SiO:; 2AlrOs'3SiOz; 3AlrOr'2SiO2; and 4AI2OB.SiOr as they were heated to 1450o C. The combined intensities of the 3.41 A and 3.38 A mullite reflections, when present' were measured at regular temperature intervals up to 1450" C. The role of crystallinity and structure of the alumina-silica components in controlling mullitization was determined by examining, both by continuous heating diffraction and by diffraction after prolonged heating at 1375' C., alumina-silica mixtures of the approximate 3Al2Oa'2SiOz mullite composition. Difiraction intensity measurements of the major mullite reflections as observed from all combinations of these six component materials after firing are plotted, as are the diffraction intensities for cristobalite and corundum.* Both peak height intensity and integrated peak area intensity measurements were recorded and found to be indicative of changes in the diffraction intensity characteristics of these minerals. Thus, the intensity of diffraction data plotted in the following illustrations representhose relative intensity changes that took place as the materials were fired. Continuous r-ray data" are based on the heating of the mixtures at a firing rate of 5.6o C. per minute. The prolonged heating data were obtained from the same mixes af ter they were held at 1375" C. Ior 24 and 72 hours respectively. All *-raying was performed using nickel-filtered copper radiation at 45 kv. and 18 ma. ExpBntlrBwtaL Drlre Efect of Alumina-Silica Mir Composition on Mullite Formation Continuous r-ray diffraction data from different mole per cents of AlrOa and SiOz in mixtures of diaspore-cristobalite while being heated to 1400o C. are shown in Fig. 1. The intensities of the (101) Iine of corun- * Three major diffraction lines attributable to the alumina component after firing have d-values of 3.48 A, 2..55 A, and 2.38 A and are representative of aalzo:. To avoid confusion with the AlzOr content of prepared mixes before firing, the mineral name for a-alzor, corundum, will hereafter be used when referring to the alumina component after firing.
1066 F. M. WAHL, R. E GRIM AND R. B. GRAF 3At2O32SiO. \ TEMPERATURE TEMPERATURE Frc. l. (Left) Intensity of diffraction by high-temperature phases in heated aluminasilica mixtures prepared from diaspore-cristobalite. cr-cristobalite; cor-corunduml Mu-Mullite. Frc. 2. (Ri.ght) fntensity of diffraction by high-temperature phases in heated aluminasilica mixtures prepared from gibbsite-silicic acid. dum, the (111) Iine of beta-cristobalite, and the combined intensities of the 3.38 A and 3.41 A mullite reflections are plotted against temperature. Mullite diffraction lines first appeared at 1200" C. from all mixes regard- Iess of their alumina to silica ratios. Just prior to the mullite nucleation temperature there was a slight but gradual increase in the intensities of the major corundum and cristobalite peaks in all mixes, but especially in
PHASE TRANSFORMATIONS IN SiOz-AtrzOt MIXTURES 1067 that with composition 3AI2Or' 2SiOz. As the temperature was raised above 1200o C., the intensity of mullite diffraction continually increased with the development of this mineral; also, there was a gradual decrease in the diffraction intensities of corundum and cristobalite. This reduction coincided with, and was inversely proportional to, mullite development' Gibbsite-silicic acid mixtures of varying composition were also examined while being heated to 1400o c. corundum and cristobalite de- COMPO S ITION Frc.3. Intensityof diffraction by high-temperature phases in heated diaspore-cristobalite mixtures after: A) t hour at 540" C.; B) 24 hours at 1375" C.; C) 72 hours at 1375' C.
1068 II. M, WAHL, R. I', GRIM AND R. B. GRAF sio2 4't :"t"r"r;"t, t:4 arzos Frc' 4. fntensity of diffraction by high-temperature phases in heated gibbsite-silicic acid mixtures after A) 24 hours at 1375" C.;B) 72 hours at 1375" C. veloped from each mixture, regardless of mole per cent variations in the mix composition (Fig. 2). rn all cases corundum was first observed at lr7o" c., but the formation temperature of beta-cristobalite ranged from 1150o to 1350" c. At no time were mullite diffraction rines detected from these components, even when heated to 1425" C. Figure 3-A shows that the diaspore in heating to 540o C. for one hour changes to corundum, and that under these conditions there is no reaction between the alumina and silica phases. rn mixtures of all compositions, corundum diffraction lines were not discernible after heating to 1375. c. (Fig. 3-B)' Mullite was present in all diaspore-cristobarite mixes after heating at 1375" c.for 24 hours (Fig.3-B) and the amount formed was directly proportional to the original alumina content of the mixture. Mixtures rich in silica retained less beta-cristobalite alter 72 hours than af.ter 24 hours (Fig. 3-c) indicating more complete mullite development with prolonged heating. All gibbsite-silicic acid mixtures deveroped mullite alter 24 hours at 1375o C., but maximum mullitization was observed from the 3Al2Og
PHASE TRANSFORMATIONS IN S\OZ'ATZOZ MIXTARES 1069.2SiOz mix (Fig. 4-A). This same development trend was maintained after 72 hours at 1375" C. (Fig. a-b). Again, less beta-cristobalite was retained alter 72 hours. Corundum development paralleled closely the amount of alumina present in the original mixes and the same was true for cristobalite development and the SiOz content. Infl,uence of Alumina Component on Mullitization Several mixes equivalent to the mullite composition (3Alrot'2SiO, were fired to determine the role of alumina structure on consequent mullite formation. Continuous diffraction data for diaspore-cristobalite, gibbsite-cristobalite, and corundum-cristobalite mixtures are shown in Fig. 5. Mullite began forming from diaspore-cristobalite at 1200" C. and Gibbsite- Cristobolite Corundum- Cristobolite 800"c looo" 1200" 1400" TE M PERATURE Frc. 5. Intensity of diffraction by high-temperature phases in heated 3AIzOr'2SiOz mixtures.
F, M. WAHL, R. E. GRIM AND R. B. GRAF Dio spore- Cri sf obolif e Corundum - Crisfobolile 54O"C (lhr) 1375'(24hr) t375"(72hr) TEMPERATURE-(TIME) Frc. 6. fntensity of diffraction by high-temperature phases in 3Alroj. 2SiO, mixtures after prolonged heating. fron' gibbsite-cristobalite at 144o" c. but none deveroped from the corundum-cristobalite mix at this rate of firing up to 1450o C. The intensity decrease of both the cristobalite and corundum peaks (Fig. 5-C) as higher temperatures are attained indicates that a possible structural readjustment to mullite is beginning to take place but is still too slight to be detected.* X-ray diffraction analyses of these same cristobalite mixtures after prolonged heating at 1375" c. showed the same trend of mullite develooment (Fig. 6). * Firing of this same mixture after extensive grinding to an extremely small particle size gave mullite at 1340" C.
PHASE TRANSFORMATIONS IN SiOz-AlzOt MIXTURES 1071 Influence of Silica Component on Mullitization continuous difiraction data from 3Al2o3.2Sioz mixtures of diasporecristobalite, diaspore-silicic acid, and diaspore-quartz while being fired to 1425" C. are listed in Fig. 7. Cristobalite and silicic acid promote mullite nucleation at 1200o and 1300o C. respectively; whereas, the quartzdiaspore mix developed mullite only after heating to 1320" C. at this rate of firing. The same reactivity trend was observed for these silica components when gibbsite was substituted for diaspore in the mixtures. The batch containing cristobalite was still the first to form mullite (Fig. 5-B)' however, none was observed from either the gibbsite-silicic acid mix (Fig. 2) or the gibbsite-quartz mix when heated to 1400o C. at this same firing rate. All three varieties of silica formed mullite with either diaspore or Diosoore - Cristobolite TEMPERATURE Frc. 7. Intensity of diffraction by high-temperature phases in heated 3AlzOg' 2SiOz mixtures.
1072 F. M. WAHL, R. E. GRIM AND R, B. GRAF Corundum- Ouorlz Gibbsite- Quorlz 54O'C(lhr.).375"(24htl 1375'(72hr.l TEMPERATURE-(TIME) Frc. 8. Intensity of diffraction by high-temperature phases in 3Alror. 2SiO2 mixtures after prolonged heating. gibbsite when heated to 1375' C. Ior 24 hours, Figures 6, 8, and 9; however' after 72 hours at this temperature, mullite difiraction intensity decreased in those batches which contained quartz as the silica component (Fig. 8). The mixtures prepared with cristobalite (Fig. 6) and with silicic acid (Fig. 9) showed the expected normal increase in mullite development after 72 hours. The effects of chemical additives on mullite nucleation from 3AlzOs '2SiOz corundum-cristobalite mixes were examined briefly. This particular mixture produced no mullite when fired at 137s" C.Ior 72 hours; however, the addition ol 5/6by weight of CaClz or BiCl3 promoted mullite nucleation from these components at 1300o and 1400' c. respectively (Fig. 10).
PEASE TRANSFORMATIONS IN SiO,-AI'Q MIXTURES 1073 DrscussroN AND CoNcLUSToNS The data show that mullitization in synthetic mixtures of various forms of alumina and silica is influenced by the structure of the components used. Limitations imposed by structure may be more important in controlling the formation of this mineral than mole per cent variations in the compositions of the alumina-silica mixes. The temperatures at which these materials combine to give mullite are thus an indirect consequence of this structure stability. The formation of mullite at 1200o C. from all diaspore-cristobalite mixes, even though they varied in composition from AIzOr'4SiOz to 4Al2O3.SiOz shows that component materials need not be present in correct molecular ratios before combination and mineral synthesis can oc- Diosoore- Silicic tcr. Gibbsite-Silicic Acid *col" Mua Corundum- Silicic Acid TEMPERATURE -(TIME) 1375'(72 hr.) Frc. 9. Intensity of diffraction by high-temperature phases in 3Alr0s' 2SiO, mixtures after prolonged heating.
ro74 F. M. WAHL, R. E. GRIM AND R. B. GRAF Corundum- Cristobolite + Co Cl2 (5U) Corundum-Cristobolite + 5% BiCls TEMPERATURE Frc. 10. rntensity of difiraction by high-temperature phases in heated 3Aleos.2sioz corundum-cristobalite mixtures treated with Ca+ and Bi++. cur. As long as the proper components are present in the mix, regardless of their relative molecular amounts, they wil combine to givelullite when heated to the proper temperature. of the varieties of alumina tested, diaspore with the composition of HAloz combines more readily with Sioz to produce mullite than does gibbsite which is AI(OH)3. On heating, each of these alumina materials ultimately converts to corundum, the diaspore at a low temperature, and gibbsite not until about 1100o c. consequent development of mullite through combination with the silica that is present is related to this preliminary corundum development. Diaspore forms corundum at a relatively low temperature (<500" c.), and contributes to mulrite development at 1200'c; whereas gibbsite does not convert to corundum until
PHASE TRANSFORMATIONS IN SiOTAIZOA MIXTURES 1075 1100'C. and does not combine with silica to produce mullite until 1440" C. The reactivity of naturally occurring corundum follows this same pattern of thermal behavior. It does not react with silica to form mullite even when held at 1375" C. for 72 hours. This suggested association between structural stability of the aluminum component and consequent mullite development is in agreement with the findings of West and Gray (1958) who showed that the reactivity in forming mullite from silica-alumina mixtures was dependent on the crystalline modification of alumina that was used. Analogous to the formation of corundum from alumina components is the transformation of the silica component to beta-cristobalite. Silicic acid, when mixed with diaspore, transforms to beta-cristobalite at 900o C. with consequent mullite nucleation at 1300" C. Quartz, however, does not form beta-cristobalite until 1200' C. and, when mixed with diaspore, does not form mullite until 1325o C. The development of mullite at 1200" C. from mixes prepared with synthetic cristobalite is as expected. The beta-cristobalite used had been prepared at approximately 900o C. from silicic acid. Above this temperature the material apparently becomes relatively unstable, more reactive, and can readily combine with the alumina component to give mullite. Because the actual temperature of mullite nucleation as determined from a heated alumina-silica mix is related to the development temperatures of corundum and beta-cristobalite, it follows that the original structures of the alumina-silica component materials, which in turn control the transformations to corundum and beta-cristobalite, are fundamental factors in limiting the ultimate development of mullite. Because mullitization generally precedes the formation of beta-cristobalite when natural alumino-silicate minerals such as kaolinite are heated, and follows beta-cristobalite development from synthetic alumina-silica oxide mixtures, it is emphasized that any material, on heating, will first tend to follow its normal transformation sequence which is determined by the structure of that material. Only after this has occurred is there a reaction to form mullite. The firing of alumina-silica components to form mullite can often be enhanced by the addition of mineralizing ions as when either Bi+# or Ca# were added to corundumcristobalite mixtures. These same ions also promoted mullitization in kaolinite (Wahl, 1958). AcruowrpnGMENTS This work forms part of a series of high-temperature r-ray diffraction studies made possible through a research grant from the National Science Foundation.
1076 F. M. WAHL, R. E, GRIM AND R. B. GRAF, RnlnReNcns ARAuAr4 S., axo Rov, R., Revised equilibrium diagram for the system AlzOrSiOz: Natotre, 184, 631-632 (1959). Bowln, N. L., aro Gnrrc, J. W., The System AlzOc-SiOz: f. Amer. Cer. Soa.,7, 238-54 (re24). Insrry, H., mvo Bwalr, R. H., Thermal behavior of the kaolin minerals: I. Res. Natl,. Bur, Stond.artls, 14, [5], 615-627 (1935). Wnsr, R. R., eno Gnav, T. J., Reactions in silica-alqmina mixtures: J. Amer. Ceram. Soc., 4r, 141, 132-136 (1958). Warr, F. M., Reactions in kaolinite-type minerals at elevated tempera.tures as investigated by continuous r-ray difiraction: Ph.D, dissertation, University of lllinois, Urbana, Ill., 87 pp. (1958). Manuscript recei,oed Actober 19, 1960.